August 10, 2025

Sun 3 bootrom souce - Tragedy with a 3-160

Yesterday I had two 3-160 boards, today I have one.

The symptoms are as follows. As soon as you apply power, all 8 DIAG LED come on immediately and stay on. Nothing on the serial port. Nothing changes when you press reset (those 8 DIAG led stay on all the time).

What happened

Here is the story. Yesterday I spent most of the day building a wooden frame to hold just this one board. I soldered up a power cable to deliver power to the P1 VME connector. I tried this with a 10 amp power supply and it did not work. The 5 volt supply got pulled down to 2.9 volts. The 3-160 required more power. Note in particular that I had 5, -12, and +12 connected from this supply.

Next I used a big 33 amp capable bench supply I have to supply power. I tried this with only +5 connected. It did not work, but I measured the 5 volt rail at several TTL chips on the board at 4.4 volts. I carefully (and in steps) increased the voltage on the Sorensen bench supply to 5.6 volts and the 3-160 began working. I saw a normal pattern on the DIAG led! I measure 4.84 volts on those TTL chips. This was yesterday.

Today I decided to continue using the bench supply with that 5.6 volt setting. I have only one wire carrying the 12 amps the board pulls, which accounts for the loss. This time I also turn on the 10 amp supply from yesterday, which is still connected to -12 and +12. I figured that the big supply can provide the 12 amps of 5 volts and the smaller supply can give us the +12 and -12.

Shortly after I turn things on, I smell the "burned electronics" smell. It is not clear just what happened.

I check voltages on the board and still measure 4.84 on Vcc.

I check the +12 and =12 at my supply connector, +12 is fine, but -12 is pulled down to 0.5 volts or something of that sort.

I move the board to a Sun pizzabox that I verify will run my other 3-160 board properly, and indeed the board is dead.

I check the 10 amp power supply that I used to give -12 and +12 and all voltages measure just fine -- 5, 12, and -12.

When the 3-160 was still connected, I measured a good +12, but the -12 was pulled down to something less that 1 volt.

What about that cable

It was wired like so: 
+5 to A32, B32, C32
+5 return to A9
+12 to C31
-12 to A31
+-12 return to C9
Closer look at the 3-160 schematics show the VME connections on page 26. It shows A31 is unconnected!

What might have gone wrong?

Is is possible that Sun expected -5 and not -12 on that supply pin? It has been mentioned that the Sun framebuffer used ECL chips that indeed require a -5 supply. I had assumed that Sun generated this by local regulation from the -12 supply. A also assumed that +12 and -12 were required for RS232 drivers. All this needs to be checked out.

The schematic shows a supply "-5TRK" that is supplied to the AM26LS29 drivers used for RS232. This comes from pins on the P3 connector that I never connected anything to. So it looks like the RS232 drivers worked from -5 to +5 not -12 to +12.

Could not supplying -5 have caused damage? Note that I found the -12 supply (supposedly not connected) was pulled down. I should check this pin on the board with a meter. Maybe they connected this n/c pin to ground.

Maybe schematics for other boards (like the 3/260) might show some details regarding power. Maybe the run serial from -5 to +12 -- I could check this with an oscilloscope.

What next?

I am optimistic that the board can be repaired, but it may be quite a challenge. But I am optimistic in general, and have no real basis for this. I could pull some socketed parts (like the rom, and eventually CPU) and see if they work in my good 3-160 board.

I am recording all this, because I intend to simply set the board aside and continue the bootrom work using my working 3-160 board, but abandon my home made frame and power supply. Someday I may find time and inclination to try to repair this board.

Feedback? Questions? Drop me a line!

Tom's Computer Info / tom@mmto.org